dimanche 25 janvier 2015

One annotation site to rule them all?

from the my-kinda-people department

I did not know that Marc Andreeson, programmer of the first graphical web browser, NCSA Mosaic, had originally included a built-in commenting feature, but rapidly disabled it when the resource implications of hosting the entire web's remarks about the entire web's content sank in (!) Fair enough. But the reason why commenting is a personal itch for me is because this is the bit of the web that so obviously represents a novel departure from traditional publishing processes. A web "page"? Meh, 2000 years off the pace, dude. Anyone anywhere can comment instantly on what's just been written? Now that is wow! like WOW! Though of course, as Gary Wolf so memorably wrote in his article about Craig's List in Wired all those years ago, the public is a motherfucker.

But from Slashdot to BMJ Rapid Responses, to Comment is Free and a slew of comments across a heap of blogs that I read on RSS, commenting is something I do, and something I follow. It seems to be where the action is. So I was very excited to come across http://hypothes.is this morning, which, as I understand it, has the ambition of taking up where Andreeson left off, and providing a reliable annotation framework for the whole web:

Digging around a bit deeper, hoping to join in, I came across this design document, written by a leading light in the project back in 2013 which outlined the approach the project is taking to facilitating the quality, and eliminating the spammy*:
The reputation of a user represents our trust of the user. In mathematical terms, we can think of the reputation as of a probability of the user telling us a correct statement. If reputation is zero then the user always gives wrong information. If reputation is 1 then the user always correct. If reputation is 0.5 then the user gives correct information in 50% of cases.

From other point of view, reputation expresses how much useful content a user has contributed. For example, in stackoverflow, more good answers I contribute, more reputation I have. Intuitively, this reputation is proportional to amount of useful work the user has done.
I was impelled to comment:
The idea of "one reputation score to rule them all" is obviously seductive when considering the ranking of multiple comments. But reputation in whose eyes? Surely a reputation system worthy of world wide use should reflect the reality that different people have different assessment of each other's reputations (particularly when we consider--ahem!--interdisciplinary contributions).
I'd certainly be interested in a system that enabled me to deprecate or appreciate comments made by other persistent identities in the knowledge that so doing would reduce or augment my probability of **me** seeing comments by that commenter in future. But I'm sure it would be a bad thing to impose my own assessment of someone else's reputation on anyone else.

But (ironically) that comment seemed not to be registered. So I'm posting it here.

FWIW: The current federated system seems reasonable. We just need better meta-information about sites that host discussion well (including archiving), and sites that cheat: pretending to host comments, but then not doing so, so we can avoid them in future. (cf SourceWatch)

* This plan involved:
Phase 1 - Simple spam triage
Phase 2 - Simple voting
Phase 3 - Metamoderation based on social graph
Phase 4 - User notifications

Update 7/2/14 1423h: 

Aucun commentaire:

Enregistrer un commentaire